Generic variation? Developments in use of generic pronouns in late 20th century spoken Danish Jensen, Torben Juel

Published in: Acta Linguistica Hafniensia: International Journal of Linguistics

DOI: 10.1080/03740460903364128

Publication date: 2009

Document version Peer reviewed version

Citation for published version (APA): th Jensen, T. J. (2009). Generic variation? Developments in use of generic pronouns in late 20 century spoken Danish. Acta Linguistica Hafniensia: International Journal of Linguistics , 41, 83-115. https://doi.org/10.1080/03740460903364128

Download date: 01. okt.. 2021

Generic variation? Developments in use of generic pronouns in late 20th century spoken Danish

Torben Juel Jensen LANCHART, University of Copenhagen, Njalsgade 136, DK 2300 Kbh. S, Denmark [email protected]

Abstract In modern Danish, a handful of pronouns may be used to refer to a generic referent. In recent decades, the second person singular pronoun du has gained ground, apparently in parallel to similar recent developments in other languages. Even though generic du may not be as old as the traditional generic pronoun man, it is not a new variant in Danish if we by “new” mean that it has come into existence within the last 30-40 years. To all appearances, it has been used before the influence from English became significant in the last part of the 20th century, and the generic du as such need therefore not be explained by contact with English, as often assumed. In order to study the spread of the generic use of du we analysed the use of pronouns with generic reference in a large sample of speakers, most of whom were recorded during sociolinguistic interviews twice – the first time in the period 1978-1989 and the second time in 2005-07. The speakers come from four locations in Denmark, and three different age cohorts are represented in the study. The results show both a rise and a decline in the use of generic du in the time interval from the 1970s and 1980s till today in accordance with the hypothesis that the increased use of generic du began in Copenhagen and started spreading from Copenhagen to the rest of the country before the time of the old recordings in the 1980s. However, the use of generic du has peaked, or is about to peak, in the Danish speech community seen as a whole, and the developments in the use of generic pronouns should probably be seen within an overall perspective of stable variation.

Keywords: generic pronouns, lifespan language change, real and apparent time studies, spatial dynamics, language variation and change

Received July 3rd 2008; revised version received September 8th 2008

1. Introduction In Danish, as well as many other languages, the second person singular pronoun may not only be used with specific reference to the addressee (as in example 1

1

below) but also to refer to an undefined person or group of persons in general, that is, with generic reference (as in example 2 below).

(1) vil du have lidt mere kaffe1 will you have little more coffee „would you like some more coffee?‟

(2) hvis man ikke bruger kondomet rigtigt så kan du få if one not use the-condom correctly then can you get børn af det men du kan også få aids af det ikke children of it but you can also get aids of it not „if you don‟t use the condom correctly you can get children but you can also get aids, right‟

The general impression, not only in Denmark but also in other European countries (and francophone Canada), seems to be that the generic use of second person pronouns is new, or that it has at least increased significantly in language use in the last decades, possibly as a result of influence from English, where the second person pronoun you has been used with generic meaning for centuries. The use of second person pronouns for generic reference has been described by a number of linguists (e.g. Bolinger 1979; Kitagawa and Lehrer 1990; Blondeau 2001; Berman 2004; Hyman 2004), but quantitative studies of variation and change in the use of second person pronouns versus other pronouns for generic reference have mostly concerned French, the first being Suzanne Laberge‟s famous study of indefinite pronouns in Montreal French (Laberge 1976; Laberge and Sankoff 1980). These studies demonstrate that the use of personal pronouns is indeed a very variable area of grammar as seen from a sociolinguistic perspective, probably, as Coveney (2003) points out, because of their close association with social relationships. It is not clear, however, whether there is actually a change going on with respect to the use of tu and vous as generic pronouns in modern French, and the issue has not, to my knowledge, been subjected to large scale studies in other languages. It has therefore been an obvious choice to include the use of pronouns with generic reference in the LANCHART study of language changes in 20th century Danish (cf. Gregersen‟s introduction to this volume). The design allows us to study the spread of the generic use of the second person singular pronoun du in the Danish speech community in relation to geography as well as gender, social class and age. The study of generic pronouns also includes linguistic factors (reference type and syntactic context) in order to assess functional

1 Unless otherwise specified all the examples in this article are original excerpts from the LANCHART corpus.

2

differences between man and du, which are probably not semantically equivalent in all respects. In addition to addressing the questions whether the use of generic du has increased in the last decades and by whom, this study highlights the more general issues of how to model and explain language variation and change, language change across the lifespan and the relationship between real and apparent time studies of linguistic variables.

2. The variable “pronoun with generic reference” Generic pronouns are traditionally counted as a subtype of the very heterogeneous class of „indefinite pronouns‟, which in addition to items expressing indefinite reference in the narrow sense such as English something, anyone and no one includes mid-scalar quantifiers (e.g. few, several, many), universal quantifiers (e.g. all, every) and identity pronouns/determiners (e.g. other, same) (Haspelmath 1997, 11-13; Dahl 2006). Characteristic of generic pronouns is that their referents are human and generalized: the descriptive reference may include the speaker, the addressee or some specific third party, but it always goes beyond that in an unspecified way (though the context of use often delimits the extension to some degree). The pronoun refers to a generalized person, and what is predicated about this referent is asserted to hold for every instantiation of the type. Using a concept coined by Langacker, the event or situation which the referent of the generic pronoun is predicated to be part of can be said to be “non-actual”: an arbitrary instance with no particular location in time or reality ‟conjured up‟ for purposes of conceiving how the world is supposed to work in general (Langacker 1997, 208). By using a generic pronoun the speaker is instructing the addressee to see the referent from a “structural” point of view: Even though it may in principle be possible to determine the pronouns precise extension, the whole point is that the predication illustrates how the world works in general, not the properties of specific persons or events.

Even though Danish has a pronoun which is predominantly used with generic reference – man, historically derived from the noun mand (≈ English man) - the pronouns used for generic reference in Danish are all polysemous having both generic and non-generic variants or uses. This is also the case in other languages such as French and English (cf. Laberge and Sankoff 1980; Kitagawa and Lehrer 1990; Ashby 1992; Steward 1995; Hyman 2004), and from a typological point of view formally distinct generic pronouns are rare in the worlds languages (Haspelmath 1997, 12). It is therefore in many cases more appropriate to work with “pronouns used with generic reference” than with generic pronouns as such, and in a quantitative analysis to count all pronouns in contexts where they are used to refer to an undefined person or group of persons in general.

3

In modern Danish, a handful of pronouns may be used with generic reference, including first person plural vi, third person singular (common gender) den and third person plural de, but except from traditional dialects only the pronouns man, second person singular du and en (derived from the numeral en ≈ English one) occur with any significant frequency. 2 En is primarily used with generic reference in contexts where it functions as the syntactic object (cf. 3 below), in a prepositional phrase (4) or as the possessive in a noun phrase (5). In these contexts en supplements man, as the latter only occurs as syntactic subject.

(3) de bad faktisk en om at vaske fingre hele tiden they requested actually one about to wash fingers whole time „in fact they asked you to wash your hands all the time‟

(4) det er nu meget sjovt synes jeg også at stå over for hh it is now very amusing think I also to stand opposite otteogtyve sproglige der der kigger undrende på twenty-eight linguistic who who look wonderingly on en når man laver fysikforsøg for dem one when one makes physics-experiment for them „it‟s rather amusing, I think, to stand in front of twenty-eight modern arts students [high school students] who look at you in wonder when you make an experiment in physics for them‟

(5) hvis ikke man kan diskutere med ens venner jamen if not one can discuss with one’s friends yes-but ved du hvad så er det jo ikke rigtigt vel know you what then is it as-you-know not right is-it „if you can‟t have a discussion with your friends well you know what, it isn‟t right then, is it‟

As the possessive the pronoun en takes the regular genitive form ens (cf. example 5 above), in other contexts it occurs as en. In the following en will always refer to the lexeme in both its forms.

Man is undoubtedly the most frequent and the most generally usable pronoun for generic reference in modern Danish. Like en it is documented as a generic

2 The pronoun man is infrequent in traditional dialects outside the isle of Zealand; in the Jutland dialect area the primary pronouns used for generic reference are en and den, in the dialect area of the isle of Funen the counterparts are de and vi (Brøndum-Nielsen 1965, V 469; Pedersen and Nyberg 1984, 61).

4

pronoun since the Middle Ages (Brøndum-Nielsen 1965, V 413-417 & 467-469), and it is the only pronoun which has generic reference as its main usage:

(6) man behøver bare at tage bussen for at høre at de unge one needs only to take the-bus for to hear that the young taler utrolig dårligt talk incredibly bad „you only need to take the bus to hear that the young people talk incredibly bad‟

(7) man kunne ikke komme ind i ungdomsklubben om aftenen one could not come in in youth-club in the-evening medmindre man blev kørt og bragt unless one was driven and brought „you couldn‟t get to the youth club in the evening unless you were brought there by car‟

Man is sometimes used with non-generic reference, though. In certain contexts it is used with first or second person meaning, that is, with specific reference to the speaker or the addressee:

(8) hvad har jeg været # ja der i midten af tresserne har what have I been yes there in the-middle of the-sixties have man været en fem seks år ikke one been a five six years not „how old was I yeah in the middle of the sixties you [I] were about five or six years, right‟

Furthermore, man is sometimes used with reference to a specific third party whose identity is unknown to the speaker:

(9) [The field worker and the informant are talking about the room in the informant‟s house in which the conversation takes place] det kunne godt ligne sådan en øh # væggen revet ned og it could well resemble such a uh the-wall torn down and så har man lagt sådan en enten en metalbjælke op eller then has one laid such a either a metal-beam up or et eller andet one or other „it looks like such a uh the wall has been torn down and then somebody has put up sort of a beam of metal or something‟

5

As already mentioned, also the second person singular pronoun du can be used with generic reference:

(10) hvis man ikke bruger kondomet rigtigt så kan du få if one not use the-condom correctly then can you get børn af det men du kan også få aids af det ikke children of it but you can also get aids of it not „if you don‟t use the condom correctly you may get children but you may also get aids, right‟

(11) vi oplever og udvikler os jo vildt forskelligt ikke we experience and develop us as-you-know wildly differently not og der dukker jo noget op nogle gange fra and there come as-you-know something up some times from barndommen eller fra tidligere som du skal have sorteret the-childhood or from earlier which you shall have sorted i ikke eller helst have sorteret i ikke in not or rather have sorted in not „we have different experiences and develop in very different ways right? and sometimes something from the childhood or from the past comes up and you have to have it sorted out right? or preferably sorted it out right?‟

The second person pronouns in the examples above do not refer specifically to the addressee but to a generalized person, as the speakers do not describe specific situations but facts about how the world is structured, according to them. Du is also used in generalizations over experiences that do not include the addressee (at least at the descriptive level):

(12) og dengang skulle man ikke nødvendigvis have and at-that-time need one not necessarily have studentereksamen for at komme ind på seminariet high school-certificate in-order to come in at teacher‟s college du kunne faktisk komme ind med med # der fra you could actually come in with with there from tredje real third real „and at that time you didn‟t need a high school certificate in order to enter a teacher‟s college you could actually be admitted from lower secondary school‟

In the example above, the addressee is not included in the reference of the pronoun du at the descriptive level, as the speaker is actually generalizing over

6

experiences that the addressee must be expected not to share, since the speaker is considerably older than the addressee, and the educational system has been changed since the time he was a student (the so-called realskole did not exist any more at the time the interviewer went to school). It is worth noticing that in (10) and (12) the speaker uses both man and du when speaking about what seems to be the same referent. This often happens in contexts where du is used with generic reference, and this phenomenon further supports the analysis that the referential potentials of du and man in modern Danish are overlapping. Du is of course also used with specific reference to the addressee (i.e. as a regular second person pronoun, cf. example 1). As it will appear later, this is, not surprisingly, its most common usage in the LANCHART corpus. Related to this use is the use of du as a discourse particle with intensifying function:

(13) ha der kan jeg huske da fik jeg altså en bagi there can I remember then got I really one in-behind du så jeg kunne mærke det ikke you so I could feel it not „At that time, I remember, I really had my bottom smacked, you bet, so I could feel it, right‟

In addition to the nominative form du, the pronoun also occurs in the accusative/oblique form dig and in the possessive forms din (singular, common gender), dit (singular, neuter) and dine (plural). In the following, du will always refer to the lexeme in all its forms.

The referential potentials of du and man are visualized in Figure 1.

Figure 1: The referential potentials of du and man

The study presented in this article only addresses the use of man and du with generic reference, that is, the intersection of the referential potentials of du and man. A precondition for studying a set of linguistic forms as expressions of a sociolinguistic variable is that the forms are “semantically equivalent” (e.g. Milroy and Gordon 2003, 185). In this respect, syntactic variables often pose problems as the “variants” are rarely completely semantically equivalent in all respects (e.g. Romaine 1984; Cheshire 1987; Lavandera 1996). This can also be said about the variable “pronoun with generic reference” in Danish as the choice between du and man in many cases seems to have significance in the speech situation. By using du the addressee is invited to see

7

the phenomena in question from the inside, so to speak, that is from the viewpoint of the generalized person, while man is more or less neutral in this respect. A likely reason for this difference between man and du is that du retains some of its second person meaning also when used generically, not in a truth conditional sense but with respect to viewpoint. This aspect has been ignored in the quantitative study which is reported in this article, but a qualitative analysis of a smaller part of the LANCHART corpus has indicated that the interpersonal potentials of du and man differ in that way (cf. Beck Nielsen, Fogtmann and Jensen, this volume). Generic du may thus differ subtly from generic man (and en), but this difference is, in my opinion, not related to the meaning of the pronouns in a truth conditional sense. The analysis of the LANCHART corpus has shown that they have the same referential potential in their generic use (cf. section 4.1), and within this envelope of variation they can therefore be seen as semantically equivalent in the sense implied by the Labovian tradition, i.e. ”truth- conditionally equivalent and used on the whole to refer to the same state of affairs” (Weiner and Labov 1983). Du and man are therefore in this study analyzed as variants of a sociolinguistic variable. This does of course not mean that the choice between the pronouns is not influenced by semantic differences of the kind hinted to above or by formal linguistic constraints, but these factors do not determine the choice of one pronoun over another. As will be demonstrated in this article, non-linguistic factors such as the origin of the speaker, her age, gender and social class influence the choice of du vs. man significantly, as does time. In my opinion, this justifies using the sociolinguistic variable as a tool in the study of change and variation in the paradigm of pronouns with generic reference in Danish.

The general impression among speakers of Danish, as it finds expression in the public debate, seems to be that the generic use of du is new in Danish and that it has spread significantly in the language use in the last decades, possibly as a result of influence from English (Jensen 2008). However, generic use of second person pronouns is rather widespread in the world‟s languages. According to Kitagawa and Lehrer „impersonal‟ use of second person pronouns is documented in Chinese, English, French, German, Gulf Arabic, Modern Hebrew, Hindi and Italian (Kitagawa and Lehrer 1990). It is furthermore documented in Spanish and Romanian (Coveney 2003, 172), Dutch (Berman 2004, 107) and Finnish (Fremer 2000; Leino and Östman 2008). According to dialect dictionaries and records, generic du is documented in the dialects of Norwegian and Swedish but allegedly only in the northern part of Scandinavia including Norway, Northern Sweden (Västerbotten and Norrbotten) and the Swedish speaking areas of Finland (NO, II 203; OF, 505; Lundeby 1996;

8

Westerberg 2004, 49).3 Generic du (tú/Þú) is also known in modern Faroese and Icelandic (FO, II 1279; IO, 1843). It is not mentioned in the dialect dictionaries of Danish, though, and according to the most comprehensive dictionary of contemporary Danish, generic du is only documented in Danish from 1973 and onwards (DDO, I 712). The extensive dictionary of modern Danish (1700-1950) does, however, mention a use of du which seems to be close to the modern generic use as seen in (10) and (11). It is described as a use “in addresses which are only apparently directed to a single person” (ODS, III 1068; ODS- supplement, III 81 [my translation]). The earliest documentation of this use comes from the play “Den 11. Junii” by Ludvig Holberg from 1723:

(14) [A (large) farmer has come to the city and is talking to two fellow farmers] Mand skulde tænke, her boede en hob riige Folk i denne Bye, thi gaar du paa Gaden, saa kand du ej komme frem for Heste, Vogne og Spraglede Laqvejer; kommer du paa Børsen, saa er der lige eens; men skal du have dine Rente-Penge, saa hedder det; Monsieur, han maa have taalmodighed; her har aldrig været saadan Trang som dette Aar (Holberg 1915 [1724], III 7)

‟one would believe that a crowd of rich people lived in this town, because if you walk down the street you cannot get through because of horses, carriages and garish footmen; if you come to the stock exchange it is the same; but if you are to collect your interest, then it goes; monsieur, he [you] must have patience, there has never been such a want as this year‟

This and other examples document that at least since the 18th century du has been used referring not only to the specific addressee but to anybody who enters the situation described. It may be argued, however, that the examples of generic du from the 18th and 19th century adduced so far are not generic to the same extent as the examples found in contemporary Danish: whereas the former in all cases include the second person, the latter are used also when generalising over situations or events which cannot be said to include the addressee (cf. example 12). But even the latter use is most likely older than from the 1970s as often assumed. In the corpus “CorDiale” comprising interviews with older, dialect speaking men and women from all parts of Denmark recorded in the 1960s and 1970s there are several examples of generic du, also examples where the addressee is not included in the reference at the descriptive level (Korpus CorDiale). The example below comes from an interview with a fisherman from

3 The dictionary of the Swedish Academy also mentions that du is used in a meaning similar to man, a use that is documented by an example of generic du used by Carl von Linné in 1734. However, this use is described as obsolete in the volume which was edited in 1925 (SAOB, VII 2303).

9

the island Falster in the south eastern part of Denmark; in this excerpt he (speaker 1) talks about how many ships there were at the place where he lived in the old days:

(15) 1 men ellers lå de hele vejen der lige hen til- du but else lay they whole way there right to you ved der hvor det hvilehjem det ligger der know there where the rest-home it lies there „but otherwise they [the ships] lay all the way to where the rest home lies, you know‟ A ja yes 1 helt langt henne quite far over-there A ja ja yes yes 1 og her ved Falstersiden der var masser af and here at the-Falster-side there were lots of dem them A ja yes 1 der var også en otte-ni-ti stykker there was also a eight-nine-ten pieces A det var alligevel en– det var- nå there was even-so it oh „even so oh‟ 1 og kom du ind til Vigsnæs der lå en hel stribe and came you in to Vigsnæs there lay a whole row også nu ser du aldrig én also now see you never one ‟and if you went to Vigsnæs there was an entire row of ships lying there too; nowadays you never see even one‟

The speaker was born in 1895, and he was at the time of the recording about 80 years old. Even though the conversation was recorded in the 1970s, examples like these indicate that du has been used also with the type of generic reference not including the addressee well before the 1970s, firstly because the speaker was rather old at the time of the recording and therefore not likely to be among the movers and shakers as regards new language changes, and secondly because he was chosen exactly because he spoke conservatively in general (the speakers

10

in the corpus CorDiale were chosen among the speakers in the local communities speaking the most genuine versions of the local dialect). To conclude, generic du does not appear to be a new variant in Danish if we by “new” mean that it has come into existence within the last 30-40 years, but this does not mean that there has not been a rise in the use of generic du within the last decades, a fact that can be documented by the data collected in LANCHART.

3. The LANCHART study of pronouns with generic reference In order to study the spread of the generic use of du we chose a sample of speakers comprising all the so called “core group” informants of the LANCHART corpus, cf. Gregersen‟s introduction to this volume. The speakers come from four locations in Denmark: the capital København (Copenhagen), the town Næstved in the southern part of Zealand, Odder in eastern Jutland and finally the village Vinderup in western Jutland. Three age cohorts are represented in the study, in the following referred to as generation 1 (speakers born 1942-63), 2 (born 1964-74) and 3 (born 1989-92). Generation 1 and 2 were recorded during sociolinguistic interviews twice – the first time in the period 1978 to 1989 and the second time in the period 2005 to 2007 - while generation 3 was only recorded once, in the period 2005 to 2007. The design allows us to study the spread of generic du in the Danish speech community in relation to geography as well as gender, social class and age (cf. Frans Gregersen‟s introduction to this volume).

If there has indeed been a rise in the use of generic du in the last decades, three hypotheses as regards the distribution of pronouns with generic reference in the LANCHART corpus can be advanced:

1. The ratio of generic du in relation to other generic pronouns changes from the first to the second recording of the informants, with du gaining ground at the expense of one or all of the other pronouns that may be used with generic reference.

2. The ratio of generic du in relation to other generic pronouns is highest in the Copenhagen data, as the change is likely to originate in the metropolitan area. This hypothesis is derived from one of the main hypotheses in the LANCHART project: the Standardization Model. It holds that linguistic changes in 20th century Danish spread from the norm centre, which is the Copenhagen area (cf. Kristiansen‟s article in this volume), and outwards to the rest of the speech community. As Næstved is more closely connected to the centre than Odder and Vinderup with respect to both geographical distance and commuting patterns, we expect the ratio of generic du in

11

relation to other generic pronouns to be higher in Næstved than in Odder and Vinderup.

3. If the increase in the use of generic du it is an ongoing language change, we would expect the proportion of du to be higher in the speech of the younger informants than in the speech of the older informants.

All occurrences of potentially generic pronouns (i.e. en, den, du and man) in the speech of the informants were tagged automatically. These occurrences were afterwards manually categorized as to the referential and syntactic aspects described below. The pronouns occurring in the speech of the interviewers were not included in the study. In the cases where the pronoun occurs in a non-completed construction (i.e. a clause which is interrupted before an interpretable intentional meaning has been expressed) the token was marked as “non-completed”. Likewise, in the cases where the same pronoun is repeated within the same construction, all the occurrences except for the last one were marked as “non-completed”. Ten percent of the occurrences in the corpus of the relevant pronouns were categorized as “non-completed”. These occurrences are not included in the study described in the following. Neither are occurrences with non-generic reference (cf. example 1, 8 and 13). The occurrences of non-generically used du and man (in “completed” constructions) constituted 64 % and 1 % of the total number of occurrences of du and man, respectively. All occurrences of generically used pronouns in completed constructions were categorized as to whether they function as “subject”, “object or predicative”, “in a prepositional phrase” or as “possessive in a noun phrase” (cf. p. 4). The results show that the latter functions of pronouns with generic reference are very rare compared to the function as syntactic subject. They also strongly suggest that en and man should not be seen as independent variants of the variable “pronoun with generic reference”. Pronouns not functioning as subjects will therefore be treated separately in the next section of this article, whereas the quantitative study described in the rest of the article only includes the pronouns du and man functioning as syntactic subjects. Included in the main quantitative study described in this article are 4392 tokens of du and 22230 tokens of man functioning as syntactic subjects and occurring in the speech of 171 different speakers in 270 different conversations. The study thus covers a considerably larger data material than similar studies of variation and change concerning generic pronouns. Laberge and Sankoff‟s study of generic pronouns in Montreal French includes 4367 tokens of “indefinite subject clitics” (on, tu and vous) in the speech of 120 speakers (Laberge and Sankoff 1980), Ashby‟s study includes 16 speakers from Tours and 587 tokens of generic on, tu and vous (Ashby 1992) and Coveney‟s study of generic

12

pronouns in Picardy French includes 30 speakers and 1031 tokens (Coveney 2003).

3.1 The paradigm of pronouns with generic reference in the corpus Other pronouns than man and du may also be used with generic reference in Danish (cf. p. 4), but analysis of the LANCHART corpus shows that they do not play any significant role in the speech of the informants, except in contexts where the pronouns do not function as syntactic subjects. In contexts where the pronouns function as subjects, other pronouns than du and man only occur in the speech of four informants from Vinderup in Western Jutland, and even here they are less frequent than man and du:4

(16) skal en så i by om æ aften sommetider så shall one then in town in the evening sometimes then bliver der altså ikke meget tid til at # til at lave becomes there really not much time to to to to make noget something „if one wants to go out in the night sometimes then there really isn‟t much time to do anything [to earn money by working]‟

In contexts where the pronouns have other syntactic functions than subject (cf. example 3-5) the pronoun en does play a significant role (see figure 2). In these contexts, the pronoun en (including the genitive form ens) comprises more than half of the occurrences of the pronouns with generic reference while forms of the pronoun du (i.e. the oblique and possessive forms dig, din, dit and dine) comprise the rest. However, as en and man seem to have complementary distribution in the language use of most of the informants (including all the informants from other locations than Vinderup) it would be wrong to treat en and man as independent variants of the variable “pronoun with generic reference”. Instead, the results indicate that en and man as generic pronouns should be analyzed as regular allomorphs in contemporary Danish, with man occurring as syntactic subject and en in other functions. This means that the paradigm of pronouns used for generic reference in the LANCHART corpus (with the exception of a few of the informants from Vinderup) is binary with the variants du and man in contexts where the pronoun function as subject and du and en in other contexts.

4 In the speech of the four informants there are 13 occurrences of en with subject function, 98 occurrences of du and 660 occurrences of man. 8 of the 13 occurrences of generic en are used by the same informant. No other pronouns occur with generic reference in the corpus included in this study.

13

Figure 2: Pronouns with generic reference

As figure 2 also shows, the contexts where pronouns with generic reference do not function as subjects are very rare: There are only 939 non- subject pronouns out of more than 27000 pronouns with generic reference. Non- subject pronouns thus comprise only 3 % of the pronouns with generic reference. It should be noted, however, that there is a rather significant difference between pronouns functioning as subjects and pronouns with other functions also with regard to the use of du: The proportion of du in the former context is 16.5 % while the proportion in the latter context is 46.3 %, almost three times larger. This difference between du vs man/en in the two contexts is statistically highly significant (χ2=556.36; 1 d.f., p<<0.01).

3.2 Change from the first to the second recording (change in real time) Several subgroups of the informants were recorded twice, namely the generation 1 informants from Copenhagen, Næstved and Odder, and the generation 2 informants from Copenhagen, Odder and Vinderup. The youngest subgroup of informants, generation 3, was only recorded once and will therefore not be included in this section. When looking at generation 1 (i.e. speakers born 1942-63) the results show that a change has indeed occurred in the use of du versus man: The proportions of du are higher in the new recordings than in the old at all three locations (see figure 3).

Figure 3: Pronouns with generic reference, generation 15

Chi square tests for independence show that the distributions of du and man are in all three locations statistically significantly different in the old compared to the new recordings (Odder: χ2=49.59, 1 d.f., p<<0.01; Næstved: χ2=6.30, 1 d.f., p<0.05; København: χ2=8.01, 1 d.f., p<0.01). As may be seen from Figure 3, the differences are largest in Odder where du comprises 8 % of the pronouns in the old recordings but 23 % in the new recordings (i.e. almost a tripling of the proportion during the 20 years interval) and smallest in Copenhagen were the proportion of du only goes up from 20 to 23 %. The results regarding generation 2 (i.e. speakers born 1964-73) are shown in Figure 4.

5 In all the diagrams in this article, the n-values indicate the total number of generically used pronouns.

14

Figure 4: Pronouns with generic reference, generation 2

Chi square tests for independence show that the distributions of du and man are statistically significantly different in the old and the new recordings in Vinderup and Odder but not in Copenhagen (Vinderup: χ2=56.18; 1 d.f., p<<0.01; Odder: χ2=8.99, 1 d.f., p<0.01; København: χ2=0.44, 1 d.f., n.s.). The proportion of generic du is stable in Copenhagen but rises considerably in Jutland. The change in the proportion of du is especially large in Vinderup where it rises from 4 % in the old recordings to 23 % in the new. It is important to notice here, however, that the old recordings from Vinderup were recorded already in 1978, that is, about ten years before the old recording from the other LANCHART communities. There is therefore almost 30 years between the old and new recordings in Vinderup whereas there is only about 20 years between the recordings in Odder and Copenhagen. To sum up, the results have confirmed the hypothesis that the du/man ratio changes from the first to the second recording of the informants, with du gaining ground at the expense of man. This indicates that there has indeed been a rise in the use of generic du in the Danish speech community within the last 2-3 decades. The results also suggest regional differences in this development in the way that the rise in use of du is biggest in Jutland and smallest in Copenhagen. These regional differences will be described further in the next section.

Even though there has been a rise in the use of du during the last decades the results clearly show that man is still the most frequently used pronoun for generic reference: In all the subgroups man comprises at least three quarters of the pronouns with generic reference, and while all informants use man both in the old and in the new recordings, some informants never use du. In the following these informants will be described as categorical man-users. When looking at the categorical man-users, the results also show that the use of du has been spreading during the period from the old to the new recordings. In the old recordings 50 informants (49 %) did not use generic du during the recordings while the number of categorical man-users (including generation 3) in the new recordings is only 10 (6 %).6 When looking at the development during the lifespan of the individual informants we see that 46 of the informants who did not use generic du in the old recording used the pronoun in this way in the new recording, while only 5 informants changed from using du to not using it. 4 informants were categorical man-users both in the old and in the new recordings.

6 The total number of informants is 103 in the old recordings and 171 in the new recordings. The number is higher in the new recordings due to the fact that the generation 3 informants were only included in the new recordings and because of 6 “supplementary” generation 1 informants in Odder and Næstved.

15

As in Sankoff and Blondeau‟s study of /r/ in Montreal French (Sankoff and Blondeau 2007) we thus see that a rather substantial part of the speakers changes their use of the variable significantly during their life span. This is remarkable seen in the light of the common assumption of relative stability in later life, after the “critical period” during puberty.

3.3 Regional differences in the use of generic du As can be seen from Figures 3 and 4 above and Figure 5 below, there are obvious regional differences as regards the use of generic du. In the old recordings, the distribution of du and man clearly confirms the hypothesis that the proportion of du is highest in the Copenhagen data and lowest in the data from Jutland, with the Næstved data in a middle position (cf. p. 11): In generation 1, the proportions of du in Odder, Næstved and Copenhagen are thus 8, 14 and 20 %, respectively (cf. Figure 3), and chi square tests for independence show that the differences between the three locations are in all cases statistically significant (Odder-Næstved: χ2=11.11, 1 d.f., p<0.01; Næstved-Copenhagen: χ2=29.45, 1 d.f., p<<0.01; Odder-Copenhagen: χ2=40.66, 1 d.f., p<<0.01). In generation 2, the proportions of du in Vinderup, Odder and Copenhagen are 4, 6 and 12 %, respectively (cf. Figure 4), and chi square tests for independence show that the differences between the two Jutland locations and Copenhagen are both statistically significant while there is no difference between Odder and Vinderup (Vinderup-Odder: χ2=1.24, 1 d.f., n.s.; Vinderup- Copenhagen: χ2=16.94, 1 d.f., p<<0.01; Odder-Copenhagen: χ2=14.57, 1 d.f., p<0,01). When looking at the new recordings, however, the results do not confirm the hypothesis. In generation 1, the proportions of du in Odder, Næstved and Copenhagen are 23, 17 and 23 %, respectively (cf. Figure 3). Chi square tests for independence show that the differences between Næstved and the two other locations are both statistically significant while there is no difference between Odder and Copenhagen (Odder-Næstved: χ2=48.17, 1 d.f., p<<0.01; Næstved- Copenhagen: χ2=42.42, 1 d.f., p<<0.01; Odder-Copenhagen: χ2=0.12, 1 d.f., n.s.). In generation 2 the proportions of du in Vinderup, Odder and Copenhagen are 23, 11 and 13 %, respectively (cf. Figure 4). Chi square tests for independence show that the differences between Vinderup and the two other locations are both statistically significant while there is no difference between Odder and Copenhagen (Vinderup-Odder: χ2=73.97, 1 d.f., p<<0.01; Odder- Copenhagen: χ2=3.19, 1 d.f., n.s.; Vinderup-Copenhagen: χ2=81.59, 1 d.f., p<<0.01).

16

The results regarding generation 3 (speakers born 1989-92) are shown in Figure 5.

Figure 5: Pronouns with generic reference, generation 3

The proportion of du in Vinderup is the double of the proportion in Næstved, and a chi square test for independence shows that the difference between Vinderup and Næstved is statistically significant (χ2=45.21, 1 d.f., p<<0.01). This is the opposite result of what would be expected from the hypothesis of generic du spreading from Copenhagen reaching Næstved before Jutland (cf. p. 11). To sum up, the results regarding the new recordings constitute a break down of the clear Copenhagen-centred pattern of the use of du which could be adduced from the old recordings from the 1970s and 1980s. Instead, the pattern in the new recordings is that the informants from Vinderup are leading as regards the use of du, followed by the informants from Odder and Copenhagen (between whom there is no difference) and with the informants from Næstved lagging behind:

1978-89: Copenhagen > Næstved > Odder / Vinderup

2005-07: Vinderup > Odder / Copenhagen > Næstved

The results are still compatible with the Standardization Model (cf. page 11), that the increased use of generic du has spread from Copenhagen to the rest of the speech community. At the time of the old recordings, the speakers from Copenhagen were clearly leaders as regards the use of generic du, and the speakers from Jutland, that is, the locations furthest away from Copenhagen, brought up the rear. What seems to have happened during the decades from the old to the new recordings is that generic du has spread across the speech community at the same time as the use of du in Copenhagen has been stabilizing. In other words, generic du has at some time between the old and the new recordings stopped spreading from Copenhagen and has become more or less evenly distributed across the country. The informants from Vinderup stand out as exceptions to this pattern as they use generic du significantly more than the speakers from the other LANCHART communities at the time of the new recordings. We have no positive explanation for this phenomenon at the moment, but it may be sought in the fact that Vinderup among the LANCHART communities is the one where the traditional dialect stands strongest, and that the pronoun man is not traditionally part of the dialect (cf. note 2). Even though the results show that the pronoun man had almost competed out the traditional generic pronouns of Jutland dialects (en and den) also in Vinderup already at the time of the old recordings (cf. the

17

previous section “The paradigm of pronouns with generic reference in the corpus”), the pronoun man may still have a weaker position in Vinderup thereby causing less resistance to the spreading of du (Jensen 2009).

3.4 Generational differences (change in apparent time) The design of the LANCHART study makes it possible to compare the use of pronouns with generic reference not only in “real” time (the language use of a given group of speakers over time) but also in “apparent” time (the language use of different age groups of speakers at a given point in time). In the old recordings, apparent time comparisons are only possible in Copenhagen and Odder, where generation 1 and 2 can be compared. In the new recordings apparent time comparisons can also be made in Vinderup and Næstved, between generation 2 and 3, and 1 and 3, respectively. In Copenhagen, the relationship between the use of generic du in generation 1 and 2 clearly is that the younger speakers use it less than the older speakers (see figure 6).

Figure 6: Pronouns with generic reference, Copenhagen

Chi square tests show that generation 1 both at the time of the old and the new recordings has a statistically significant higher proportion of du than generation 2 (old recordings: χ2=51.40, 1 d.f., p<<0.01; new recordings: χ2=95.47, 1 d.f., p<<0.01). A somewhat similar pattern can be seen in the results from Odder (see figure 7).

Figure 7: Pronouns with generic reference, Odder

Chi square tests show that generation 1 at the time of the new recordings has a statistically significant higher proportion of du than generation 2, while there is no difference at the time of the old recordings (old recordings: χ2=0.79, 1 d.f., n.s.; new recordings: χ2=59.95, 1. d.f., p<<0.01). In contrast to Copenhagen, there is thus no difference between the two generations at the time of the old recordings in Odder, but in the new recordings there is. What has happened is that the speakers from generation 1 have increased their use of du more than generation 2 in such a way that a Copenhagen-style distribution pattern has been imported at the time of the new recordings (there is no difference in the use of generic du between Odder and Copenhagen informants neither in generation 1 nor 2 at the time of the new recordings, cf. p. 17). In Næstved, generation 2 is not represented, but generation 1 can be compared to generation 3 at the time of the new recordings (see figure 8).

18

Figure 8: Pronouns with generic reference, Næstved

A chi square test shows that generation 1 has a statistically significant higher proportion of du than generation 3 (χ2=136.84, 1 d.f., p<<0.01). The pattern is therefore the same as in Copenhagen and Odder: The younger speakers have a lower proportion of du than the speakers from generation 1. In Vinderup, generation 1 is not represented, but generation 2 can be compared to generation 3 at the time of the new recordings:

Figure 9: Pronouns with generic reference, Vinderup

A chi square test shows that generation 2 has a statistically significant higher proportion of du than generation 3 (χ2=63.14, 1 d.f., p<<0.01). The relationship between generation 2 and 3 thus seems to be the same as between generation 1 and 2/3 in the other locations: Again, the younger speakers have a lower proportion of du than the older speakers. To conclude, the results clearly do not confirm the hypothesis that the proportion of generic du is higher among the younger than the older speakers. On the contrary, the use of generic du seems to peak in generation 1, while the younger generations use it less - at least as regards the new recordings but also in the old recordings from Copenhagen. The data from Vinderup indicate that this relationship between the older and the younger speakers also exists between generation 2 and 3.

At the time of the old recordings, the apparent time comparisons in Odder and Copenhagen indicate that the use of generic du is stable (as regards Odder) or in decline (as regards Copenhagen). This will make us predict that the ratios of generic du in relation to other pronouns with generic reference in the speech communities seen as wholes are about the same in the new and the old recordings in Odder and lower in the new than in the old recordings in Copenhagen. As only generation 1 and 2 are represented in the study at these locations it is not possible to confirm or disconfirm the prediction directly in real time. To do this, we would need a representative sample of the speakers in Odder and Copenhagen with respect to age also at the time of the new recordings. To have that, we would definitely need to include in the new recordings a group of informants younger than generation 2 (who where 33-42 years old at that time). Because no new (and younger) informants were included in the new recordings, the mean age of the speakers is necessarily about 20 year older in the new compared to the old recordings. However, as the real time comparisons over the lifespan of the panel speakers show that both generation 1 (in both Odder and Copenhagen) and 2 (in Odder) have actually increased their use of du during the

19

20 years period from the old to the new recordings (cf. Figure 3 and Figure 4) it is somewhat unlikely that the predictions based on apparent time results are accurate, at least as regards Odder. It seems more likely that the use of generic du has actually increased in the Odder speech community seen as a whole. In the new recordings, the pattern of the use of generic du is clearly that generation 1 uses the pronoun generically the most and generation 3 uses it the least. The results thus strongly suggest that the rise in the use of generic du has peaked – or is close to doing so - in the Danish speech community seen as a whole, and that the use of generic du in the future will decline. Having in mind that the apparent time method is based on the assumption that the language use of the individual speakers remains more or less stable after they have left adolescence (e.g. Bailey 2004, 320), while the study of the generic du has demonstrated a significant change (a rise) over time in both generation 1 (in Copenhagen, Næstved and Odder) and generation 2 (in Odder and Vinderup) we might want to make a reservation with regard to this prediction. However, taking into consideration that the use of generic du in Copenhagen has only increased relatively little (generation 1) or has been stable (generation 2), and that the rise in the use of generic du seems to have spread from Copenhagen to the rest of the Danish speech community, it seems fair to conclude that the use of generic du has peaked, or is about to peak, in the Danish speech community seen as a whole. Even though the use of du may still be rising in generation 1 and in generation 2 (outside Copenhagen), the increasing population of younger speakers with a lower use of generic du will make the proportion of du decline over time. The developments in the use of generic pronouns should therefore probably be seen within an overall perspective of stable variation.

An alternative analysis of the results is that the use of generic du has not changed in the Danish speech community over time but that it is “age graded”, meaning correlated with a particular phase in life and repeated in successive generations (e.g. Bailey 2004, 324). In this case, the age grading would be that the use of generic du was correlated with the life phase represented by the informants in generation 1 at the time of the new recordings, that is, 45-65 years. This would mean that the reason that the informants in generation 1 and 2 generally increase their use of generic du in the period between the old and the new recordings is simply that they get older. This type of age grading would also explain why generation 1 uses generic du more than generation 2, and generation 2 more than generation 3, though it could not explain the regional differences. The best way to test the hypothesis of age grading is to compare two groups of speakers with the same age and from the same location but recorded at two different points in time (i.e. a trend study). If the differences regarding the use of generic du described above are due to age grading, there should be no difference in the proportions of du between the two sets of recordings.

20

The LANCHART design makes such a comparison possible only with regards to one of the LANCHART speech communities, namely Vinderup, where generation 2 in the old recordings from 1978 has the same age as generation 3 in the recordings from 2006-07 (15-17 years). As shown in Figure 4 and Figure 5, respectively, the proportion of du in generation 2 in the old recordings is 4 % while the proportion in generation 3 in the new recordings is 12 %. This difference is statistically significant (χ2=16.58, 1. d.f., p<<0.01), i.e. the 15-17-year-olds in 1978 have significantly fewer occurrences of generic du than the young people of the same age in 2006-07. It can therefore be refuted that the development in the use of du in Vinderup is due to grading alone. In Odder and Copenhagen a comparison is only possible between subgroups of the informants in generation 1 and generation 2.7 In Odder, the subgroup of generation 1 informants who at the time of the old recordings had the same age as generation 2 in the new recordings (35-45 years) consists of only 6 informants, so a comparison does not really make sense here. In Copenhagen, however, there are 12 informants in the subgroup of generation 1 overlapping in age at the time of the recording with generation 2. The result of the comparison is shown in Figure 10.

Figure 10: Pronouns with generic reference, age 35-45 Copenhagen

A chi square test shows that the differences between the two sets of informants are statistically significant (χ2=47.43, 1 d.f., p<<0.01), and the comparison demonstrates that there has been a decline in the use of generic du in the period 1987-2006 in the age group 35-45 years. Even though the direction of change is the opposite as in Vinderup, the result of the comparison of similar age groups in the old and new recordings from Copenhagen is also incompatible with the hypothesis of age grading. Instead, the results both as regards Vinderup and Copenhagen further support the analysis that there has been an increased use of generic du spreading from Copenhagen to the rest of the country starting before or around the time of the first recordings in the 1970s and 1980s, and that the use of du at least in Copenhagen peaked some time between the old and the new recordings.

3.5 Gender and social class The results of the study show that gender and social class influence the use of pronouns with generic reference. In the LANCHART corpus seen as a whole men use generic du significantly more than women, in the old as well as in the new recordings. As regards social class the results show that in the corpus seen

7 In Næstved, not even such a comparison is possible as there is no age overlap between the informants of generation 1 in the old recordings and generation 3 in the new recordings.

21

as a whole “working class” speakers use generic du more than “middle class“ speakers in the old as well as in the new recordings. Space does not allow going further in to these issues in this article, however interesting they may be from a sociolinguistic viewpoint. They will be described and discussed in detail in a forthcoming article.

4. Functional differences between du and man It has been established above that the variable “pronoun with generic reference” is influenced by time as well as the age and origin of the speaker, but this does of course not entail that the choice between du and man is not influenced by linguistic factors too. In the following, two important factors will be described.

4.1 Reference The factor ”reference” concerns the scope of the reference of pronoun in the actual context in which it is used. This factor has been included in the study due to the hypothesis that du is dispreferred when the addressee is excluded from the reference of the pronoun at the descriptive level (type 2 and 4 below) because it retains some of the deictic (second person) meaning also when used generically. This hypothesis is inspired by Kitagawa and Lehrer who make the following claim about the impersonal you in English:

Impersonal use of a personal pronoun cannot exclude in its reference what its normal (deictic) use would signify (e.g., the addressee cannot be excluded from the reference of impersonal you by a phrase such as ‟I don‟t mean you personally‟). (Kitagawa and Lehrer 1990, 742)

All occurrences of du and man with generic reference have therefore been categorized in four groups according to their type of reference:

1. Everybody or a group of people not further defined but including both speaker and addressee (see also example 5, 6, 10 and 11): du kender godt det # at hvis man er glad for den you know well this that if one is happy with the-one man har # så er man altid ked af at de skal one have then is one allways sorry of that they shall skifte change „you know what it‟s like if you‟re happy with the one [your children‟s teacher] you have you‟ll always be sorry that they [your children] have to change‟

22

2. A group of people including the speaker, but excluding the addressee (see also 3, 4, 7 and 12): så gik vi ind over markerne og øh så maste man then walked we in over the-fields and uh then struggled one det tog en time at gå derop så # så kommer man it took an hour to walk up-there then then come one derop ikke og så får man at vide at skolen er up-there right and then get one to know that the-school is lukket closed „[if there was a snowstorm] then we walked across the fields and uh you had to struggle it took an hour to walk up there then you get up there right and then you‟re told that the school is closed [because of the snow]‟

3. A group of people including the addressee, but excluding the speaker: men du sagde du gik i skole i Saksild # til at starte med but you said you went in school in Saksild to to begin with og # var det så efter syvende klasse man skulle ind til and was it then after seventh grade one should in to Odder eller hvad Odder or what „but you said that you went to school in Saksild in the beginning and was it then after the seventh grade that you had to go to Odder‟

4. A group of people excluding both the speaker and the addressee: og jeg kørte på et toholdsskift og han var inde ved and I drove on a two-team-shift and he was in by militæret og vi så aldrig hinanden der var næsten som the-military and we saw never each-other there was almost as at være som hvis man er sømandsenke det vil det altså to be as if one is sailor‟s-widow it will it det er vel det samme it is probably the same „and I was working in double shift and he was in military service and we never saw each other it was almost as if you are a sailor‟s widow that‟s the same I guess‟

In the overwhelming majority of cases it was fairly straightforward to categorize the pronouns into one of the four types above, even though the analyzers

23

sometimes needed to take into consideration a rather large textual context in order to disambiguate the reference. In a few cases, though, it was not possible from the context to determine the reference, and the tokens where then given more than one code (e.g. type 1 AND 2). These occurrences are excluded from the results described below.

Du is represented in all the categories of reference (with proportions ranging from 15 to 22 %) documenting that the referential potential of du in contemporary Danish is the same as that of man - as regards generic use. The type of reference definitely does not determine the choice of du versus man, though it does seem to influence it (see below). Figure 11 shows the results concerning the influence of reference on the choice of pronouns in the new recordings in the LANCHART corpus seen as a whole. Only the informants who actually use generic du during the recordings are included in this part of the study (i.e. the non-categorical informants, cf. p. 15), because the “choice” of pronoun cannot be influenced by its reference if the informant always uses man.

Figure 11: Non-categorical speakers, new recordings

The results support the hypothesis that du is dispreferred when the addressee is not included in the reference: The two categories “group including speaker but excluding addressee” and “group excluding both speaker and addressee” taken together (= “group excluding addressee”) have a statistically lower proportion of du than the categories “group including both speaker and addressee” and “group including addressee but excluding speaker” taken together (= “group including addressee”) (χ2=28.07, 1 d.f., p<<0.01).8 The difference between the proportions of du in the two united categories is only four percentage point, though, and the results more than anything else show that speakers actually very often use generic du also in situations where the addressee is excluded from reference. As can be seen from the n-values in figure 11, “group excluding addressee” is the most frequent type of reference for generic du in the corpus – as it is for man. In other words: The speakers do not seem to have any problems using du even though the reference of the pronoun does not include the addressee.

The expansion of the use of generic du in the period between the old and the new recordings might implicate that also the impact of the factor “reference” has

8 As the use with reference to “a group of people including the addressee, but excluding the speaker” for both pronouns is extremely rare compared to the other generic uses of the pronouns it does not influence the overall result. There are only 85 occurrences of pronouns with this type of reference.

24

changed from the old and the new recordings. However, the results show that this is not the case (see figure 12).

Figure 12: Reference type of generic du’s, non-categorical speakers

The proportion of the generic du‟s occurring in contexts where the addressee is excluded from the reference (type 2 and 4 above) is 64 % in the old recordings and 62 % in the new recordings, but a chi square test show that the distributions of reference types of the generic du in the speech of the non-categorical speakers in the old and the new recordings are not statistically significantly different (χ2=0.77, 1.d.f., n.s.).

4.2 Syntactic context (occurrence in conditional constructions) The factor ”syntactic context” was included in the analysis primarily because Laberge and Sankoff (1980) in their study of Canadian French found that ”implicative constructions” (statements of cause and effect) favour the generic use of second person pronouns instead of the traditional generic pronoun on. According to Laberge and Sankoff, the reason for this is that implicative constructions are in themselves indicators of generality. Because of their “hypothetical nature” they ”work to diminish the possibility of ambiguity with the second person referent when tu (or vous) is used”, thus favouring the use of tu/vous instead of on (Laberge and Sankoff 1980, 280). Ashby‟s study of Tours French shows similar results (Ashby 1992), and Leino and Östman argues that the Finnish generic second person construction “involves some degree of hypotheticality or displacement from the here-and-now” (Leino and Östman 2008, 47). This applies to French (and Finnish), but an obvious hypothesis is that it also applies to Danish. The pronouns with generic reference occurring in the corpus were therefore categorized as to whether they occur in a conditional construction9 or not. A conditional construction was defined as a clause complex in which a clause (the protasis) specifies hypothetical, general or uncertain

9 Laberge and Sankoff‟s results concern what they call ”implicative constructions” which are defined as ”…statements of cause and effect. They consist of two sentences, the first of which (the protasis) sets up a supposition and the second (the apodosis) states the implications” (Laberge and Sankoff 1980, 277). Like in the study presented in this article, the syntactic relation between the two sentences need not be one of embedding; they may also be coordinate or simply juxtaposed. Even though this definition of implicative construction is very similar to the definition of conditional constructions presented in this article, I have chosen the term “conditional” as it may be more narrow than “implicative” as used by Laberge and Sankoff. Ashby uses the term “conditional structures” (Ashby 1992, 145).

25

circumstances on which the actualization of another clause (the apodosis) is asserted to depend. Examples of pronouns with generic reference occurring in conditional constructions are seen in 2, 4, 5, 7, 10, 14, 15 and 16. Example 5 is repeated below for convenience:

(5) hvis ikke man kan diskutere med ens venner jamen if not one can discuss with one‟s friends yes-but ved du hvad så er det jo ikke rigtigt vel know you what then is it as-you-know not right is-it „if you can‟t have a discussion with your friends well you know what, it isn‟t right then, is it‟

Our hypothesis was that a conditional construction is a favourable environment for du, because the risk of the addressee misinterpreting the pronoun as referring to her is minimized – due to the hypothetical or generalizing nature of this construction. We therefore expected the proportion of du to be larger in conditional constructions than in other constructions seen as a whole. The results as regards the non-categorical speakers in the new recordings are shown in Figure 13.

Figure 13: Non-categorical speakers, new recordings

A chi square test for independence shows that the factor ”syntactic context” has a statistically significant influence on the choice of du versus man (χ2=160.23; 1 d.f.; p<<0.01). The proportion of du is almost 60 % larger in conditional constructions than in other syntactic contexts seen as a whole. It is noteworthy that conditional constructions are a very frequent context for pronouns with generic reference in general: 23 % of all the pronouns with generic reference in the corpus occur in conditional constructions, and 31 % of the occurrences of generic du. The results clearly support the hypothesis that conditional constructions favour the use of du. However, this does not necessarily indicate that generic du is somehow attracted to conditional constructions; a more likely reason for the result is that the use of generic du is precluded in some contexts, namely in contexts where the risk of misinterpreting the pronoun as referring specifically to the second person is too big. In (17) below, for example, it will be difficult if not impossible to use du instead of man for that reason. With du as subject, the utterance will most likely by understood as a direct request or suggestion to the addressee, presumably due to the modal verb burde in past tense (marking irrealis):

26

(17) man burde sætte nogle tiltag i værk for at you/one ought-to launch some initiatives in doing in-order to # brande dialekterne lidt bedre # brand the-dialects a-little better „you ought to launch some initiatives in order to brand the dialects a little better‟

A conditional construction is a type of context where the risk of misinterpreting the pronoun as referring specifically to the second person is reduced, and this is probably the reason for the relatively higher proportion of du. There is no change with respect to the impact of the factor “syntactic context” between the old and the new recordings: The proportion of generic du occurring in conditional constructions is 31 % also in the speech of the non-cate- gorical informants in the old recordings.

5. Conclusion The LANCHART study of pronouns with generic reference has confirmed the assumption that there was a rise in the use of generic du in late 20th century Danish, though the pattern in the use of generic pronouns in the speech community is more complex than might be expected. The results document an increasing use of generic du in real time within the period 1978-89 to 2005-05 in generation 1 (speakers born in the period 1942- 63) in Odder, Næstved and Copenhagen and in generation 2 (born in the period 1964-73) in Vinderup and Odder. These generations of speakers have thus changed their use of generic pronouns significantly during their life span. The results therefore query the apparent time method in the study of language change, as this method is based on the assumption that the language use of individual speakers remains more or less stable after they have left adolescence. The results indicate that the increased use of du has spread from Copenhagen to the rest of the country: At the time of the old recordings, the speakers from Copenhagen were clearly leaders as regards the use of generic du, while speakers from Jutland (i.e. the locations furthest away from Copenhagen) used it the least, and with the speakers from Næstved in a middle position as regards language use as well as geography. What seems to have happened, then, during the decades from the old to the new recordings is that the increased use of generic du has spread across the speech community at the same time as the use of du in Copenhagen has been stabilizing or even declining. This is supported by the fact that there is no change in the use of du in the youngest group of speakers in Copenhagen, generation 2, and that a comparison between two groups of speakers of age 35-45 years in Copenhagen reveals a decline in the proportion of generic du 1987-88 to 2005-06. In contrast, a similar comparison of young

27

speakers (15-17 years) in Vinderup demonstrates a rise in the proportion of du from 1978 to 2006-07. In other words, generic du has at some time between the old and the new recordings stopped spreading from Copenhagen and has become more or less evenly distributed across the country. The results of the study supports the Standardization Model which holds that changes in language use in 20th century Danish spread from the norm centre, which is the Copenhagen area, and outwards to the rest of the speech community. Comparison of the different age groups indicates clearly that the use of generic du has peaked somewhere between the time of the old recordings and the new recordings and is now declining in the Danish speech community seen as a whole. The use of generic du thus seems to peak with generation 1, who were 45-65 years old at the time of the new recordings, while the younger generations use it less - at least as regards the new recordings but also already in the old recordings from Copenhagen. The data from Vinderup show that this relationship between the older and the younger speakers also holds between generation 2 and 3 (born in the period 1989-92). To sum up, the results show both a rise and a decline in the use of generic du in the time interval from the 1970s and 1980s till today in accordance with the hypothesis that the increased use of generic du begun in Copenhagen and started spreading from Copenhagen to the rest of the country before the time of the old recordings in the 1980s. However, the use of generic du has peaked, or is about to peak, in the Danish speech community seen as a whole. Even though the use of du may still be rising in generation 1 and in generation 2 (outside Copenhagen), the increasing population of younger speakers with a lower use of generic du can be expected to make the proportion of du decline over time. When data from generation 3 in Odder and Copenhagen become available for analysis, we expect the proportions of du to be lower in generation 3 than in generation 2, and the proportion of du in Copenhagen to be equal to or lower than the proportions in the rest of the country.

The rise in the use of generic du documented by the study does not seem to be directly connected to a change in the usage potential of du. At least the study has not found such a change, as the pronoun du is represented in the same categories of generic reference in the old recordings as in the new ones. In contrast, the study has documented that du already at the time of the old recordings was used as a generic pronoun on a par with the traditional generic pronoun man. If there has been a reanalysis of the pronoun du expanding its referential potential from specific second person singular reference to second person singular reference as well as generic reference, this reanalysis happened before the time of the old recordings. In fact, it is fair to assume that du has been used with generic reference (also with reference excluding the addressee) well before the 1970‟ies as it can be documented among old speakers of traditional dialects already in the

28

1960s and 1970s. Generic du is therefore definitely not a new variant of pronouns with generic reference in Danish if we by “new” mean that it has come into existence within the last 30-40 years. To all appearances, it has been used before the influence from English became significant in the second half of the 20th century, and generic du as such needs therefore not be explained by contact with English. Furthermore, generic use of du is documented in North Scandinavian dialects already in the 18th century (Westerberg 2004, 49) and generic use of second person pronouns occurs in languages such as Chinese and Arabic (Kitagawa and Lehrer 1990) where influence from English is not an obvious explanation. It seems more likely, then, that the possibility of abstracting from the deictic domain (the role-identity of the addressee in the speech situation) to generic reference is a cross-linguistic potential which has been exploited in Danish more or less independently of the development in English. The increase in the generic use of the second person pronoun in Danish – and arguably also in other languages such as (Canadian) French, Spanish, Norwegian, Swedish and Finnish (Thibault 1991; Lundeby 1996; Fremer 2000; Fernández 2007; Leino and Östman 2008) - within the last decades may, of course, be caused by, or supported by, increased influence from English. However, this is not the only possible explanation. As regards the use of tu/vous for generic reference in French, language internal causes have been proposed by a number of researchers: Because on in much everyday French is used as the normal first person plural pronoun, it has become “functionally overloaded”. This makes some speakers use tu/vous for generic reference instead of on in order to minimize ambiguity. This behaviour then in turn causes a reorganisation of the pronoun system (Laberge 1976; Laberge and Sankoff 1980; Blondeau 2001; Coveney 2003). Leaving aside that an explanation like this only pushes the explanation problem one step back (why have speakers begun using on instead of nous?), it is important to consider whether such language internal explanations could be advanced for the use of du instead of man in Danish. With respect to the use of du with generic reference in subject function, it is not obvious that the change should be caused be some other change in the pronoun system, as man cannot be said to be “functionally overloaded” in Danish. Man does have other functions than generic reference (cf. example 8 and 9), but these uses are very infrequent compared to its use as generic pronoun. Ambiguity between generic and non- generic reference is thus unlikely to be a problem in cases where the intended reference is generic. On the contrary, ambiguity problems would more likely occur when using du with intended generic reference as this pronoun is most often used with specific reference, i.e. to the addressee. The reality of such ambiguity problems is rendered probable by the results of the study showing that du is significantly favoured by conditional constructions, a type of context in

29

which the risk of the addressee misinterpreting the pronoun as referring to her, is minimized. However, in the cases where the pronoun has other functions than syntactic subject, changes in the pronoun system may play a role. The results show that the proportion of du is considerably higher when the pronoun with generic reference does not function as subject than when it functions as subject. This difference may be caused by the fact that the pronoun en (which supplements man in these contexts) has gone out of use as a generic pronoun in contemporary Danish with the very exception of these relatively rare contexts where a pronoun with generic reference does not function as syntactic subject. This may give the pronoun en a weaker position than subject man causing less resistance to the spreading of du. The most likely explanations for the development in the use of generic du should be sought in the interplay between a subtle meaning difference between du and man and general societal changes with respect to linguistic interaction during the last 3-4 decades. Even though man and du may both be used with generic meaning, there seems to be a connotation with du which man does not have: Du signals a recipient-oriented viewpoint with respect to what is described (c.f. Berman 2004). By using du the speaker invites the addressee to see the phenomenon in question from the inside, so to speak, that is from the viewpoint of the generalized person, while man is more or less neutral in this respect. A likely reason for this difference between man and du is that du retains some of its deictic meaning also when used generically, not in a truth conditional sense but with respect to viewpoint or orientation. A qualitative study of the use of pronouns with generic reference among Copenhagen generation 1 informants (i.e. the speakers who appear to be the leaders as regards the increased use of generic du) (cf. Beck Nielsen, Fogtmann and Jensen, this volume) indicates that du is very often used when the speaker wants to involve the addressee: Du more often than man occurs during “assessment actions” and in passages characterized by “enactment” serving as illustrations or exemplifications of more general claims. This difference in the pragmatics of du and man may be the key to understanding the increase in the use of du in the last decades demonstrated by the LANCHART study. However, this of course depends on whether we can find societal changes in the way Danes interact with each other that have made the potential of du more relevant than before. In Beck Nielsen, Fogtmann and Jensen (this volume) the trends towards intimization and personification, which have often been argued to be characteristic of late-modernity discourse, are advanced as likely causes for the increase in the use of generic du. The present study does, however, not indicate that the rise in the use of generic du starting before the time of the old recordings is the beginning of a language change in the way of du replacing man as generic pronoun in Danish.

30

On the contrary, the study indicates when that the developments in the use of du should be seen within an overall perspective of stable variation.

References "Korpus Cordiale". http://cordiale.dyndns.org/. Ashby, W. J. 1992. "The Variable Use of on Versus Tu/Vous for Indefinite Reference in Spoken French". French Language Studies 2: 135-157. Bailey, G. 2004. "Real and Apparent Time". In The Handbook of Language Variation and Change, edited by J. K. Chambers, Peter Trudgill and Natalie Schilling-Estes. Malden, MA: Blackwell Publishing. Berman, R. A. 2004. "Introduction: Developing Discourse Stance in Different Text Types and Languages". Journal of pragmatics 37: 105-124. Blondeau, H. 2001. "Real-Time Changes in the Paradigm of Personal Pronouns in Montreal French". Journal of Sociolinguistics 5, no. 4: 453-474. Bolinger, D. 1979. "To Catch a Metaphor: You as Norm". American Speech 54: 194-209. Brøndum-Nielsen, J. 1965. Gammeldansk Grammatik I Sproghistorisk Fremstilling. København: Københavns Universitets Fond til Tilvejebringelse af Læremidler. Cheshire, J. 1987. "Syntactic Variation, the Linguistic Variable and Sociolinguistic Theory". Linguistics 25, no. 2: 257-282. Coveney, A. 2003. "'Anything You Can Do, Tu Can Do Better': Tu and Vous as Substitutes for Indefinite on in French". Journal of Sociolinguistics 7, no. 2: 164-191. Dahl, Ö. 2006. "Indefinite Pronouns". In Encyclopedia of Language and Linguistic, vol. 5, edited by Keith Brown. Oxford: Elsevier. Fernández, S. S. 2007. "To Konkurrerende Generaliserende Konstruktioner På Spansk". Ny Forskning i Grammatik 14: 67-87. Fremer, M. 2000. "Ve E Du Då - Generisk Du Hos Ungdomar Och Vuxna Talare". In Ungdom, Språk Og Identitet, edited by Ulla-Britt Kotsinas, Anna-Brita Stenström and Eli-Marie Drange. København: Nordisk Ministerråd. Haspelmath, M. 1997. Indefinite Pronouns. Oxford: Oxford University Press. Holberg, L. 1915 [1724]. "Den 11. Junii". In Ludvig Holdbergs Samlede Skrifter, vol. 3, edited by Carl S. Petersen. København: Gyldendalske Boghandel. Hyman, E. 2004. "The Indefinite You". English Studies 85, no. 2: 161-176. Jensen, T. J. 2008. "Om Du I Betydningen ‟Man‟". Nyt fra Sprognævnet 2008, no. 3: 1-8. Jensen, T. J. 2009. "Standardisering, Globalisering Og Lokalisering - Generiske Pronominer I Vinderup". In I Mund Og Bog, edited by Henrik

31

Hovmark, Iben Stampe Sletten and Asgerd Gudiksen. København: Københavns Universitet. Kitagawa, C. and A. Lehrer. 1990. "Impersonal Uses of Personal Pronouns". Journal of Pragmatics 14: 739-759. Laberge, S. 1976. "The Changing Distribution of Indefinte Pronouns in Discourse". In Language Use and the Uses of Language, edited by Roger W. Shuy and Anna Shnukal. Washington: Georgetown University Press. Laberge, S. and G. Sankoff. 1980. "Anything You Can Do". In The Social Life of Language, edited by Gillian Sankoff. Philadelphia: University of Pennsylvania Press. Langacker, R. 1997. "Generics and Habituals". In On Conditionals Again, edited by Angeliki Athanasiadou and René Dirven. Amsterdam: John Benjamins Publishing Company. Lavandera, B. 1996. "Where Does the Sociolinguistic Variable Stop?". In Towards a Critical Sociolinguistics, edited by Rajendra Singh. Amsterdam: Benjamins. Leino, P. and J.-O. Östman. 2008. "Language Change, Variability, and Functional Load. Finnish Genericity from a Constructional Point of View". In Constructional Reorganization, edited by Jaakko Leino. Amsterdam/Philadelphia: John Benjamins Publishing Company. Lundeby, E. 1996. "Om Man Og En Og Du". In Språket Lever - Festskrift Til Margareta Westman, edited by Svenska Språknämnden. Stockholm: Svenska Språknämnden. Milroy, L. and M. Gordon. 2003. Sociolinguistics - Method and Interpretation. Malden, MA: Blackwell Publ. Pedersen, K. M. and M. Nyberg. 1984. "Når Bønder Fortæller". Danske Folkemål 26: 1-246. Romaine, S. 1984. "On the Problem of Syntactic Variation and Pragmatic Meaning in Sociolinguistic Theory". Folia Linguistica XVIII: 409-437. Sankoff, G. and H. Blondeau. 2007. "Language Change across the Lifespan: /R/ in Montreal French". Language 83, no. 3: 560-588. Steward, M. M. 1995. "Personally Speaking...Or Not? The Strategic Value of on in Face-to-Face Negotiation". French Language Studies 5: 203-223. Thibault, P. 1991. "La Langue En Mouvement: Simplification, Régularisation, Restructuration". LINX: Linguistique Institut Nanterre-Paris-X 25: 79- 92. Weiner, J. E. and W. Labov. 1983. "Constraints on the Agentless Passive". Journal of Linguistics 19: 29-58. Westerberg, A. 2004. Norsjömålet under 150 År. Uppsala: Kungl. Gustav Adolfs Akademien för svensk folkkultur.

32

Ddo. 2003-2005. Den danske ordbog. Edited by Ebba Hjorth, Kjeld Kristensen, Henrik Lorentzen and Lars Trap-Jensen. København: Det Danske Sprog- og Litteraturselskab, Gyldendal. Fo. 1998. Føroysk Orðabók. Edited by Jóhan Hendrik W. Poulsen, Marjun Simonsen, Jógvan i Lon Jacobsen, Anfinnur Johansen and Zakaris Svabo Hansen. Tórshavn: Føroya Fródskaparfelag & Fródskaparsetur Føroya. Io. 2002. Íslensk Orðabók. Edited by Mörður Árnason. Reykjavik: Edda. No. 1966-. Norsk ordbok - Ordbok over det norske folkemålet og det nynorske skriftmålet. Edited by Alf Hellevik. Oslo: Det Norske Samlaget. Ods. 1918-1956. Ordbog over det danske Sprog. Edited by Verner Dahlerup, Lis Jacobsen and H. Juul-Jensen. København: Det Danske Sprog- og Litteraturselskab. Ods-S. 1992-2005. Ordbog over det Danske Sprog - Supplement. Edited by Anne Duekilde and Sv. Eegholm-Pedersen. København: Det Danske Sprog- og Litteraturselskab. Of. 1982-. Ordbok över Finlands svenska folkmål. Edited by Olav Ahlbäck. Helsingfors: Forskningscentralen för de inhemska språken. Saob. 1893-. Svenska Akademiens Ordbok öfver Svenska Språket. Lund: Svenska Akademien.

33